Performance Evaluation of Deep Learning Models for Breast Cancer Classification
Wajahat Akbar, Abdullah Abdullah, Sajid Ahmed Ghanghro, Muhammad Inam Ul Haq, Mohib Ullah · 2023
Globally, breast carcinoma is the most prevalent type of cancer, causing almost nine hundred thousand deaths yearly. It is possible to reduce the fatality rate if the disease is detected at an early stage and appropriately diagnosed. Early diagnosis may prevent it from dissemination and prevent premature casualties from contracting it. Breast cancer (BC) researchers encounter several challenges when trying to distinguish benign from malignant tumours and when trying to draw conclusions about mild and advanced cancer. Through the use of machine learning algorithms, all tumours can be identified using algorithms that can locate and recognize patterns. Most patients with breast cancer die from improper diagnosis and treatment every year. Deep learning algorithms have proven quite effective in the detection of breast cancer in recent years. However, there remains considerable scope for improving these techniques. Despite considerable advancements, the use of deep learning methods, specifically in the context of machine learning, can further enhance their efficiency. In this paper, we compared three different models on the BreakHis dataset, which is openly available, and we demonstrated effecienctv2 accurate classification among all. It achieved an accuracy of 99% and 98% for training and validation, respectively.